Re: [Help-glpk] Multiple LP problem solving

[Dorian Tourin-Lebret]

2011/6/8 Andrew Makhorin <address@hidden>

> I'm evaluating GLPK in my C code to solve a multiple LP problem. I'm
>  trying to minimize C.x - d, where C is a real n x m matrix, d is a
>  known n-dimensional vector and x is the unknown m-dimensional vector
>  I'd like to get. The constraints deal with bounds for every variable.

> I understand I can minimize one _expression_ like z = a.x1 + b.x2 + ...
> (documentation p14), but how should I write the n expressions:

> z1 = a1.x1 + b1.x2 + ...
> z2 = a2.x1 + b2.x2 + ...
> z3 = a3.x1 + b3.x2 + ...
> ...

> so that the variables x1, x2, ... are shares by every _expression_?

Your question is unclear. Do you mean that you have several objective
functions within one lp, or that you have several lp's with the same
coefficient matrix, or what?

I mean I have several objective functions within one LP. I'm trying to find the values of the variables x1, x2, ..., xN that minimize the following functions:

z1 = a1.x1 + b1.x2 + ...
z2 = a2.x1 + b2.x2 + ...
z3 = a3.x1 + b3.x2 + ...

The variables are common to each function, but the coefficient are not. I'm basically trying to minimize "Cx - d", given a coefficient matrix C and a constant vector d.

Btw, I'm used to working with MATLAB and the lsqlin() function to solve that kind of LP.

Am I clear enough?